Optimization of an Advanced Hybrid Wing Body Concept ......Optimization of an Advanced Hybrid Wing Body Concept using HCDstruct Version 1.2 Jesse R. Quinlan and Frank H. Gerny NASA

Optimization of an Advanced Hybrid Wing Body

Concept using HCDstruct Version 1.2

Jesse R. Quinlan∗ and Frank H. Gern†

NASA Langley Research Center, Hampton, VA, 23681

Hybrid Wing Body (HWB) aircraft concepts continue to be promising candidates forachieving the simultaneous fuel consumption and noise reduction goals set forth by NASA’sEnvironmentally Responsible Aviation (ERA) project. In order to evaluate the projectedbenefits, improvements in structural analysis at the conceptual design level were neces-sary; thus, NASA researchers developed the Hybrid wing body Conceptual Design andstructural optimization (HCDstruct) tool to perform aeroservoelastic structural optimiza-tions of advanced HWB concepts. In this paper, the authors present substantial updatesto the HCDstruct tool and related analysis, including: the addition of four inboard andeight outboard control surfaces and two all-movable tail/rudder assemblies, providing afull aeroservoelastic analysis capability; the implementation of asymmetric load cases forstructural sizing applications; and a methodology for minimizing control surface actuationpower using NASTRAN SOL 200 and HCDstruct’s aeroservoelastic finite-element model(FEM).

Nomenclature

AATT = Advanced Aircraft Transport TechnologyANGLEA = Angle of attack degree of freedomANN = Artificial Neural NetworkAOA = Angle of attack in degreesARMD = Aeronautics Research Mission DirectorateBWB = Blended Wing BodyCFD = Computational Fluid DynamicsDLM = Doublet-Lattice MethodEL = Deflection of elevator in degreesERA = Environmentally Responsible Aviation Project at NASAFEM = Finite Element ModelFLOPS = Flight Optimization SystemGA = Genetic AlgorithmHCDstruct = Hybrid wing body Conceptual Design and structural optimizationHWB = Hybrid Wing BodyIN = Deflection angle of inner elevon N in degreesLaRC = Langley Research CenterLHS = Latin Hypercube SamplingMDOPT = Multidisciplinary Design OptimizationNASA = National Aeronautics and Space AdministrationND8 = NASA D8ON = Deflection of outer elevon N in degreesOpenVSP = Open Vehicle Sketch PadOREIO = Boeing Open Rotor Engine Integration on a BWBPDF = Probability Density Function

∗Aerospace Engineer, Aeronautics Systems Analysis Branch, 1 N Dryden Street, and AIAA Member.†Assistant Branch Head, Vehicle Analysis Branch, 1 N Dryden Street, and AIAA Member.

1 of 13

American Institute of Aeronautics and Astronautics

PSI = Pounds-per-square-inchRN = Deflection of rudder N in degreesTE = Trailing EdgeUI = User InterfaceURDDN = Degree of freedom N

I. Introduction

Hybrid wing body (HWB) or Blended Wing Body (BWB) aircraft concepts are considered promisingalternatives to conventional tube and wing configurations due to their large potential fuel savings and

increased aerodynamic efficiency.1 Fuel burn reductions of 25% and higher have been published for someultra-high capacity configurations carrying up to 800 passengers. Most notable is the work of Liebeck andhis co-workers at The Boeing Company. Their 450 passenger BWB-4501L design resulted in the 8.5% scaleX-48B flight demonstrator shown in Fig. 1.2

Figure 1. Boeing’s X-48B Blended Wing Bodyflight demonstrator (NASA Photo).

A significant difficulty in dealing with HWB design opti-mization has always been the lack of a data base of knownflying designs which may serve as calibration and validationpoints for conceptual design and optimization programs likeFlight Optimization System (FLOPS),3 especially when com-pared to the vast amount of available tube and wing aircraftdata. When transitioning from the conceptual to the prelimi-nary design phase, the aircraft designer needs to be sure thatthe design chosen for further optimization is actually a viabledesign, and as a result of the lack of validation cases, the devel-opment of improved fidelity analysis tools becomes imperativefor the conceptual design loop.

To validate the projected fuel burn and noise reduction po-tential of HWB designs for NASA’s Environmentally Respon-sible Aviation (ERA) project, significant efforts have been put forward to develop advanced structural andaerodynamic analysis tools for HWB conceptual design optimization. Aerodynamic methods improvementhas been geared towards increased fidelity in-the-loop methods like enhanced panel codes and computationalfluid dynamics (CFD).4 Recently developed structures tools include finite element model (FEM) based anal-yses to provide enhanced capabilities for HWB centerbody sizing and weight estimation which recently led tothe approval for public release of the HCDstruct tool (Hybrid wing body Conceptual Design and structuralanalysis).5,6

The current paper first describes updates to HCDstruct reflected in the latest release, hereafter referredto as HCDstruct Version 1.2 (V1.2). Specifically, the additional aeroservoelastic analysis capabilities aredescribed as related to the recent inclusion of four inboard and eight outboard elevons and two all-movabletail or rudder assemblies. Next, several asymmetric load and maneuver cases recently added to the struc-tural optimization analysis are described, and related results are presented using the Boeing Open RotorEngine Integration on a BWB (OREIO) concept. Finally, work toward optimizing maneuver control surfacedeflection schedules for HWB vehicles with multiple redundant control surfaces is presented.

II. HCDstruct Version 1.2 Development

Following years of development and several applications4–10 of earlier tool versions utilizing a symmetrichalf-model formulation, HCDstruct V1.2 has recently been approved for release by NASA Langley ResearchCenter (LaRC) as a substantial update to the original software. This latest version is based on a fullwing-tip-to-wing-tip model formulation that also includes numerous control surface configuration options.Additionally, the original set of load cases used to constrain the structural optimization has been expandedto now include asymmetric maneuvers. The release of HCDstruct V1.2 also includes complete input andoutput decks for the Boeing OREIO HWB concept, and in contrast to the original version, the latest softwareversion is deployed as a stand-alone, Windows executable requiring no additional licensed software.

Detailed overviews of HCDstruct V1.0 and V1.1 are available in the literature;4,6,7 thus, in this section,

2 of 13


only new details of HCDstruct V1.2 are discussed. In the first section, II.A, newly-available control surfaceconfiguration options available in HCDstruct V1.2 are presented. Next, in section II.B, the asymmetricmaneuver load cases recently added to the HCDstruct V1.2 optimization analysis are described. Finally, insection II.C, structural optimization results for the Boeing OREIO are compared to those of the previousrelease, with particular emphasis on the increased fidelity of the aft body sizing due to the inclusion ofasymmetric loadings.

II.A. Aeroservoelastic Capability

In addition to implementing and verifying the full wing-tip-to-wing-tip model formulation and the refineduser-interface (UI), the primary development for HCDstruct V1.2 consisted of configuring and verifyinginboard and outboard wing and tail control surfaces. The Boeing OREIO configuration, for which an OpenVehicle Sketch Pad (OpenVSP)11 model and full-model HCDstruct FEM are shown in Fig. 2, was usedas the baseline configuration in support of this developmental work. In Fig. 2, the HCDstruct FEM alsoincludes the control surface coordinate systems, systems weights as concentrated masses, and rigid barsused to support the propulsion systems and vertical tails. In addition to the central elevons (elevators),the as-designed Boeing OREIO also included eight outboard wing elevons, four inboard wing elevons, twoall-movable vertical tails, two rudder surfaces on the vertical tails, and high-lift Kreuger flaps at the leadingedges of the wings.

Figure 2. OpenVSP model of the Boeing OREIO concept (left) and HCDstruct V1.2 finite-element model of the sameBoeing OREIO configuration (right).

In HCDstruct V1.0, the only control surface modeled was the single starboard-side elevator, as this wasthe only surface required to trim the symmetric half-model in pitch. In the current version of the software, theinboard and outboard wing elevons and the all-movable tails and rudders were implemented and configuredfor use with both the symmetric half-model FEM and also the full wing-tip-to-wing-tip FEM formulation.The leading edge flaps included in the as-designed OREIO configuration were not modeled for this effortsince they are of little utility in the selected critical load cases used to constrain the structural optimization.Thus, the half-model now supports up to eight independent control surfaces and the full model up to 16independent control surfaces, noting that the user may only select either of the all-movable tail or rudder.The control surfaces available in HCDstruct V1.2 are highlighted on the FEM model of the OREIO in Fig. 3,where the outboard elevons are shown in green, the inboard elevons are shown in yellow, the central elevonsare shown in red, the rudders are shown in blue, and the all-movable tails shown in orange. In its currentimplementation, these control surfaces are aerodynamically modeled, only. That is, the control surfaces arecomprised of Doublet-Lattice Method (DLM) panel elements for which the aerodynamic forces are splinedto the wing rear spar.

The control surfaces can be configured either by direct linking (via NASTRAN12 AELINK specifications),where the surface deflections are linked to those of the two central elevons, or by specifying individualdeflections directly (via NASTRAN TRIM specifications). For both cases, the user can provide AELINKcoefficients or surface deflections, respectively, in the HCDstruct executive input deck. These additionalcontrol surface configuration options now permit the application of asymmetric maneuvers for the full wing-tip-to-wing-tip model, which are discussed in the following section.

3 of 13


Figure 3. Control surfaces configured and available for use in HCDstruct V1.2 shown overlaid on the top-view of theBoeing OREIO HCDstruct V1.2 finite-element model.

II.B. Asymmetric Load Cases

As was suggested in previous studies,6,10 significant uncertainty associated with predictions of the structuralweight for advanced HWB concepts is likely attributed to the aft body assembly. For HWB concepts like theOREIO, the aft body must be designed to carry the inertial loads of the propulsion systems, vertical tails,and skins and also the bending loads associated with the central elevon hinge moments and any loads on theaft body due to asymmetric maneuvers or engine-out conditions. While the original version of HCDstruct didsize the aft body structure for the inertial loads of the vertical tails and propulsion systems, only symmetricload cases were available for the analysis.

Another major development to HCDstruct for V1.2 was the implementation of asymmetric load cases,in addition to the four symmetric load cases available in V1.0. These additional asymmetric load casesinclude a 1.0-G dynamic overswing maneuver and a 1.0-G rudder reversal maneuver. For these asymmetricmaneuvers, the full wing-tip-to-wing-tip model is constrained rigidly at the nose grid point in the streamwisedirection only; all other translational and rotational degrees of freedom are constrained using reactionaryforces and moments via SUPORT1 NASTRAN specifications. A summary of these additional asymmetriccases are shown in Table 1, in addition to those of the original symmetric load cases, where ANGLEA refersto the angle of attack; URDD2 refers to the lateral acceleration; URDD5 refers to the pitch acceleration;and URDD6 refers to the yaw acceleration. The non-free aeroelastic degrees of freedom (AESTAT) specifiedon the TRIM cards for the asymmetric load cases include: SIDES (sideslip angle), ROLL (roll rate), PITCH(pitch rate), YAW (yaw rate), URDD3 (vertical acceleration), and URDD4 (roll acceleration). The fixedaeroelastic degrees of freedom specified on the TRIM cards for the symmetric load cases include those ofthe asymmetric loads and URDD2, URDD5, and URDD6. Additionally, for dynamic overswing and rudderreversal load cases, the rudder control surface deflections and sideslip angles are specified according to Ref. 13.

Table 1. Summary of load cases used by HCDstruct for structural optimization of HWB aircraft concepts.

Case Class Description Free Trim Variables Availability

1 Symmetric 2.5-G Limit Load ANGLEA V1.0+

2 Symmetric -1.0-G Limit Load ANGLEA V1.0+

3 Symmetric 2.0-G Taxi Bump ANGLEA V1.0+

4 Symmetric 2P Cabin Overpressurization ANGLEA V1.0+

5 Asymmetric 1.0-G Dynamic Overswing ANGLEA, URDD: 2, 5, 6 V1.2+

6 Asymmetric 1.0-G Rudder Reversal ANGLEA, URDD: 2, 5, 6 V1.2+

4 of 13


II.C. Boeing OREIO Results

Figure 4. Component and total structural weight convergence history for Boeing OREIO optimization using HCDstructV1.2.

The Boeing OREIO concept was structurally optimized using HCDstruct V1.2 with the control surfaceconfiguration and load cases described in the previous section, and the results of this optimization arecompared to those of previous studies.7,10 The weight convergence history is shown in Fig. 4 for the presentanalysis, demonstrating relative convergence of the total structural weight. In Fig. 5, the component-leveland total structural weight predictions for the OREIO using HCDstruct V1.2 are compared to those ofprevious studies using HCDstruct V1.0, FLOPS,3 and Boeing’s multidisciplinary optimization (MDOPT)approach.14 As compared to the results of HCDstruct V1.0, the aggregate effect of all the HCDstructV1.2 developments was to drive the structural weight predictions lower and nearer to the Boeing MDOPTresults, which represent the highest-fidelity optimization analysis referenced in this work. Specifically, theaft body structural weight prediction agrees significantly better with that of Boeing at about 56% differentas compared to about 136% different using HCDstruct V1.0. While the wing structural weight lessened byabout 8% as compared to HCDstruct V1.0, the total structural weight prediction for HCDstruct V1.2 agreeswell with the results of Boeing. As was also found previously,10 the results compared less favorably to thoseof FLOPS, which is unsurprising due to the lack of relevant design data available for HWB concepts requiredto tune the reduced order weight estimation models used by FLOPS.

Aeroservoelastic analysis results for the dynamic overswing and rudder reversal load cases are shown inFigs. 6 and 7, respectively. For each figure, the optimized model displacement results are shown overlaidwith contours of von Mises stress distributions for top, side, and bottom views of the Boeing OREIO,demonstrating the asymmetric loadings captured by these maneuver cases. Based on these results, the rudderreversal load case appears to stress the wing mid-sections considerably more than the dynamic overswingmaneuver, while both cases place significant stress on the aft body at the points of vertical tail attachment.The load asymmetry is further apparent by inspecting the relative wing displacements for both cases.

III. Control Surface Deflection Schedule Optimization

One of the many benefits of HWB designs utilizing numerous trailing edge (TE) control surfaces is theinherent control redundancy. With numerous inboard and outboard elevons, for example, in addition tocentrally-placed elevators, the same maneuver may be achieved with numerous control surface deflectionschedules. Each of these theoretical control surface deflection schedules requires a specific actuation power,which can be formulated as an optimization problem for a specified maneuver due to the duplicity of thesystem. With the additional control surfaces configured for use in HCDstruct V1.2, the tool now provides a

5 of 13


Figure 5. Comparisons of component-level and total structural weights for the Boeing OREIO concept for HCDstructV1.2, HCDstruct V1.0, Boeing MDOPT,14 and FLOPS3 (from left to right).

Top View

Side View

Bottom View

Von Mises Stress[PSI]

Figure 6. Dynamic overswing load case results for the Boeing OREIO, where translational displacements are shownoverlaid by von Mises stress contours in pounds-per-square-inch (PSI) for top, side, and bottom view, from left toright, respectively, of the optimized structural model.

6 of 13


Top View

Side View

Bottom View

Von Mises Stress[PSI]

Figure 7. Rudder reversal load case results for the Boeing OREIO, where translational displacements are shownoverlaid by von Mises stress contours in pounds-per-square-inch (PSI) for top, side, and bottom view, from left toright, respectively, of the optimized structural model.

platform for performing aeroservoelastic analyses in support of such an optimization effort. In the followingsection, III.A, a methodology for performing an optimization on the total actuation power requirements fora given maneuver is described, and in section III.B, the methodology is applied to the Boeing OREIO tooptimize the actuation power required to perform a 2.5-G pull-up maneuver.

III.A. Methodology

Significant work toward using artificial neural networks (ANNs) and genetic algorithms for minimization ofactuation power for HWB concepts utilizing multiple redundant TE control surfaces has been performedby researchers at Virginia Polytechnic University (VaTech) through a NASA Aeronautics Research MissionDirectorate (ARMD) Seedling Fund Project.8,15 By building a surrogate model for total required controlsurface hinge moment (as a proxy for actuation power) using an ANN, and combining this model with agenetic algorithm (GA) optimization, the researchers have performed computationally-efficient optimizationsof total hinge-moment for a 2.5-G pull-up maneuver for the Boeing OREIO. This work has demonstratedthe great potential for actuation power savings, and in fact, the latest results suggested potential actuationpower savings of over 30% as compared to using the central elevators alone to perform the maneuver. Inthis paper, this optimization problem is approached using a more conventional gradient-based optimizationmethod to serve as a point of comparison for the ANN and GA optimization method. To this end, the multi-disciplinary gradient-based SOL 200 optimizer within NASTRAN is used to implement this methodology,and details of this approach are described below.

Due to the high-dimensionality of the optimization problem, a database of control surface deflectionschedules was first constructed. Latin hypercube sampling (LHS)16 was used to populate this database of1000 deflection schedules with all deflections bounded by -45°and +45°and modeled as normal distributionsaround a mean of 0°. The port side deflections for each schedule mirrored those of the starboard side dueto the symmetry of the target maneuver. For each of these database samples, a static aeroelastic analysis

7 of 13


was performed using NASTRAN SOL 144 for a symmetric 2.5-G pull-up maneuver, and the sum of theabsolute hinge moments for all the control surfaces was computed according to Eq. 1, where φ is the sum ofthe absolute hinge moments, Hi, and serves as the objective function for subsequent optimizations. Uponcompleting the SOL 144 analyses, the corresponding objective function results represented a baseline measureof the total hinge moments (the proxy for actuation power) required to perform the maneuver.

φ =∑i

|Hi| (1)

With a baseline data set for the objective function, the next step was to perform the gradient-basedoptimization on the total hinge moment and to identify the deflection schedule corresponding to the minimumobjective function. Using the database of 1000 control surface deflection schedules as initial conditions,NASTRAN SOL 200 was configured to perform an optimization on Eq. 1. Leveraging the symmetry ofthe maneuver by linking the port side surfaces to those of the starboard side using AELINK cards, eightdesign variables required specification using DESVAR cards, including the starboard side elevator, rudder,and inboard and outboard elevons. These eight design variables were linked to aerodynamic monitor points,MONPNT1, using DVPSURF cards, for which monitor points were configured to compute the control surfacehinge moment for each of the 16 control surfaces. Design responses for each of these control surfaces weredefined using DRESP1 cards, which were then used in conjunction with a DEQATIN card for Eq. 1 and acomposite DRESP2 response card to compute the objective function for each design cycle of SOL 200.

A Python script was written to control the entire optimization analysis, including constructing the LHSdatabase; configuring all the SOL 144 and SOL 200 NASTRAN run files; spawning five instantiations ofNASTRAN simultaneously; and post-processing the NASTRAN output files to compile the resultant trimvariables and control surface hinge moment data. Using a local computing cluster, a complete SOL 200analysis for the LHS database required approximately 21 wall clock hours. The results of this analysis forthe Boeing OREIO are presented in the following section.

III.B. Results for the Boeing OREIO

For this study, five separate analyses were performed in order to both establish baseline measures for opti-mization results comparisons and to also investigate the effect of design variable constraints on the optimizedsolutions. These five analyses are described here and are summarized in Table 2. The baseline analysis case,BASELINE, consisted of trimming the aeroelastic model for a 2.5G maneuver using only the elevator controlsurfaces, with the only free variable being the elevator deflection angle. The total hinge moment required forthis BASELINE case provides a measure of the conventional control approach used for fixed wing aircraft.The second analysis case, MINLHS144, refers specifically to the solution corresponding to the minimumtotal hinge moment solution in the database of SOL 144 static aeroelastic 2.5G maneuver solutions. CaseMINLHS144 does not represent an optimization; rather, it is simply the minimum total hinge moment solu-tion from the set of 1000 LHS randomized samples. The remaining analysis cases, 300LHS200, 375LHS200,and 450LHS200, correspond to the minimum total hinge moment solutions in databases of SOL 200 opti-mized aeroelastic 2.5G maneuver solutions subject to control surface deflection constraints of 30.0°, 37.5°,and 45.0°, respectively. For SOL 200 cases, the case identifier consists of the deflection constraint, the ini-tial condition approach, and the solution sequence. For example, case 300LHS200 refers to the optimumconfiguration from the solution set defined by a 30.0°deflection constraint (300 ), performed using the LHSdatabase as initial conditions (LHS ) and using solution sequence 200 (200 ).

Also included in Table 2 are the hinge moment figures and the net reduction in required hinge momentas compared to the BASELINE case. Based on these figures, a significant opportunity for reducing thetotal hinge moment required to perform a 2.5G maneuver exists as compared to the conventional elevator-only approach. As the available design space was expanded by increasing the control surface deflectionconstraints, SOL 200 found increasingly optimal solutions. With a 30.0°constraint, reductions of 11.6% inrequired total hinge moment were found, and as this constraint was increased to 45.0°, this reduction increasedto approximately 42.9%. Based on these results alone, there exists considerable opportunity for reducingactuation power by leveraging redundant control surfaces in conjunction with the aeroelastic flexibility ofthe aircraft.

With the large LHS database of control surface deflection schedules used for case MINLHS144 and thecorresponding optimizations performed for 300LHS200, 375LHS200, and 450LHS200, these grouped analysesprovide a substantial database of total hinge moment results that help to provide some insight into the

8 of 13


Table 2. Summary of hinge-moment optimizations using SOL 200 and for baseline cases.

Case ID Solution Independent Design Deflection Hinge Reduction from

Sequence Variables Constraint Moment [in-lbs] Baseline [%]

BASELINE SOL 144 N/A None 3.497E+6 0.0

MINLHS144 SOL 144 N/A 90.0° 3.094E+6 11.5

300LHS200 SOL 200 8 30.0° 3.093E+6 11.6

375LHS200 SOL 200 8 37.5° 2.666E+6 23.8

450LHS200 SOL 200 8 45.0° 1.996E+6 42.9

solution state space. This data is presented in the form of probability density functions (PDFs) in Fig. 8 tobetter highlight the aggregate trends present in these results. Inspecting these curves, one can immediatelysee the multi-dimensionality and complexity of the design space that when paired with the gradient-basedoptimization algorithm used in SOL 200 results in multi-modal distributions for each set of SOL 200 analyses.Further, the net effect of relaxing the design variable constraints can be seen by inspecting the 300LHS200,375LHS200, and 450LHS200 analyses in the red (dashed line), green (dash-dot line), and purple (long-dashline) curves, respectively. As the design constraint is relaxed from 30.0°to 45.0°, the PDF peaks shift towardlower values of the objective function, and the PDFs exhibit higher levels of variance. Furthermore, whencomparing these three PDF curves to that of the SOL 144 LHS database solutions in black (solid line), onecan see that the SOL 200 optimization algorithm finds significantly more-optimal solutions for every LHSdatabase sample. The median and variance of the SOL 200 PDFs are significantly lower than that of theSOL 144 PDF.

Additional to the PDF curves, the minimum solutions for each analysis data set are annotated in Fig. 8using the case identifiers shown in Table 2. These data points for cases MINLHS144, 300LHS200, 375LHS200,and 450LHS200 are shown in purple (diamond), green (gradient), red (delta), and black (circle), respectively,and demonstrate the relative reduction in total required hinge moment as compared to the BASELINE caseshown in blue (square). By using a 30.0°deflection constraint, the SOL 200 optimized solution falls nearlyidentically on the minimum solution in the set of SOL 144 solutions for the randomized LHS database, whileboth are still considerably lower than that of the BASELINE solution. By increasing the deflection constraintfrom 30.0°to 37.5°and further to 45.0°, the reduction in total hinge moment for the optimal solution is mademore significant, where cases 300LHS200, 375LHS200, and 450LHS200 show reductions of over 4.04E+05,8.31E+05, and 1.50E+06 in-lbs, respectively, or 11.6%, 23.8%, and 42.9%, respectively, as compared to theBASELINE case. These results demonstrate the significant reductions in required hinge moment achievableby leveraging the natural flexibility of the wing and the control surface redundancies along the TE to achievethe lift profile required for a trimmed 2.5-G maneuver.

Table 3. Summary of control surface deflections for baseline and optimized cases.

Case ID AOA [°] EL [°] I1 [°] I2 [°] O1 [°] O2 [°] O3 [°] O4 [°] R1 [°]

BASELINE 6.690 13.900 0.000 0.000 0.000 0.000 0.000 0.000 0.000

MINLHS144 7.350 3.360 -9.210 -20.800 41.500 89.600 5.960 0.375 -0.729

300LHS200 7.727 12.400 -11.561 -19.514 29.969 29.981 4.408 -7.721 -0.780

375LHS200 7.734 9.170 -10.797 -20.263 36.358 36.167 19.602 -6.371 9.165

450LHS200 7.866 3.960 -9.373 -21.288 45.000 45.000 45.000 3.341 3.961

As the optimizations leverage the flexibility of the airframe structure, the specific schedule of controlsurface deflections used to perform the 2.5G maneuver may change drastically. In this work, the onlyconstraints on the optimization are the minimum and maximum deflection bounds; the airframe FEM usedhere was the weight-optimized model of the OREIO discussed earlier. Hence, some additional insight may behad by inspecting how the optimizer achieves the reductions in total required hinge moment via the deflection

9 of 13


Figure 8. PDF representations of the objective function in Eq. 1 for the SOL 144 analyses (solid black line) and forthe SOL 200 optimization analyses (dashed red line). The minimum solutions for the SOL 144 and SOL 200 results arerepresented by the black circle and red delta, respectively, and the baseline elevator-only configuration data point isshown in the blue square.

schedules. The actual deflections computed for cases BASELINE, MINLHS144, 300LHS200, 375LHS200, and450LHS200 are summarized in Table 3 and are presented graphically in Figs. 9 and 10. In Fig. 9, the controlsurface deflections are shown for cases BASELINE and MINLHS144 for the elevator (EL), the inboard elevons(I1, I2), the outboard elevons (O1, O2, O3, O4), and the rudder (R1). Only the starboard control surfacesare shown, since the port side solution is necessarily identical due to the symmetry of the problem. The angleof attack (AOA) is also plotted. In Fig. 10, the same data is presented for cases 300LHS200, 375LHS200,and 450LHS200.

Comparisons of Figs. 9 and 10 and inspection of Table 3 serve to highlight the wide differences in solutioncontrol surface deflection schedules used to attain a trimmed 2.5G maneuver for these analyses. In Fig. 9,case MINLHS144 shows a stark contrast to that of case BASELINE, for which all control surface deflectionsbesides EL are zero and the AOA is near 6.7°. For case MINLHS144, the AOA increases slightly while the ELdecreases slightly, and the remaining control surfaces deviate strongly from zero. In fact, the outboard elevonsO1 and O2 are nearly 41.5°and 89.6°, respectively, while the remaining control surfaces exhibit deflectionsranging from -20.8°to 5.96°. For cases 300LHS200, 375LHS200, and 450LHS200, the deflections are generallyas varied; though, as the deflection constraint is increased from 30.0°to 45.0°, for example, surfaces O4 andEL change direction while surface deflections for I1, I2, O1, O2, and O3 all generally increase in magnitude.Additionally, as the deflection constraint is increased, the AOA also increases slightly. From this data,it is clear that the converged control surface deflection schedule for minimal total required hinge momentis representative of the complex interplay between control surface size and configuration and the specificaerodynamic and structural modes of the model present in the aeroelastic analysis. It is within this contextthat one must consider the present computational results; additional considerations such as aircraft safetyfeatures, human pilot factors, and limitations of the on-board controls and actuation systems will placeadditional constraints on the minimum required total hinge moment or control actuation power.

10 of 13


Figure 9. Column representations of the trimmed angle of attack (AOA), elevator deflection (EL), inboard elevondeflections (I1, I2), outboard elevon deflections (O1, O2, O3, O4), and rudder deflection (R1) for the elevator-onlybaseline case (BASELINE) and for the minimum LHS static solution case (MINLHS144), shown from left to right,respectively.

Figure 10. Column representations of the trimmed angle of attack (AOA), elevator deflection (EL), inboard elevondeflections (I1, I2), outboard elevon deflections (O1, O2, O3, O4), and rudder deflection (R1) for the SOL 200 optimizedcases 300LHS200, 375LHS200, and 450LHS200, shown from left to right, respectively.

IV. Summary and Future Work

The current paper first presents recent additions to the HCDstruct tool, including advanced aeroservoe-lastic analysis capabilities and asymmetric load cases, all of which are included in the release of HCDstructV1.2. For verification and tractability purposes, airframe structural optimization results for the Boeing OR-EIO configuration were compared to those of previous HCDstruct V1.0 and to Boeing MDOPT and FLOPSanalyses and were found to agree better with the results of Boeing, overall. HCDstruct now supports theconfiguration and use of up to 16 independent control surfaces comprising a completely actuated trailingedge, and these control surfaces are aerodynamically modeled, only. However, in future releases, these control

11 of 13


surfaces will also include structural model representations.Using in part the newly-implemented control surfaces, a gradient-based optimization methodology for

minimizing actuation power requirements for an HWB concept with a high degree of control surface redun-dancy was developed. This methodology was developed to support a joint effort with VaTech toward the useof ANN and GA optimization of control actuation power for a HWB, where the methodology presented hereserved as a baseline optimization case using conventional gradient-based methods. This approach requiredbuilding a LHS database of randomized control surface deflections, which were then used to perform staticaeroservoelastic analyses using NASTRAN SOL 144 and subsequently used as initial conditions for optimiza-tions on the sum of the absolute hinge moments for all of the control surfaces using NASTRAN SOL 200.Results of this analysis suggested potential reductions in total hinge moment required to perform a 2.5-Gpull-up maneuver for the OREIO by nearly 43% as compared to using only the central elevators. Similaroptimizations for additional maneuver cases are being pursued. Planned work also includes extending thismethodology to mission performance applications to better demonstrate the viability of the optimal controlsurface deflection schedules identified using this approach.

The next release of HCDstruct (Version 2.0) will comprise a major update, including, among other mod-ifications, a new user interface and the addition of a generalized tube and wing aircraft modeling capability.Furthermore, the next version will support the analysis of advanced fuselage designs like the double-bubbleconcept, which was implemented in support of the NASA D8 (ND8) systems analysis and design studycurrently underway at NASA LaRC under the Advanced Aircraft Transport Technology (AATT) project.Additionally, structural model representations of the control surfaces are being developed, and an aerody-namic matching algorithm is currently being implemented in order to match the DLM aerodynamics withavailable CFD data to perform flutter analysis at transonic Mach numbers.

Acknowledgments

The authors thank Craig Nickol, Rakesh Kapania, Moustaine Adegbindin, Nathan Love, and JosephSchetz for numerous helpful discussions. This work was supported by NASA’s ERA and AATT projects,with additional support from the NASA Pathways Program.

References

1Liebeck, R. H., “Design of the BWB Subsonic Transport,” 40th AIAA Aerospace Sciences Meeting and Exhibit , Reno,NV, January 2002.

2Risch, T., Cosentino, G., Regan, C. D., Kisska, M., and Princen, N., “X-48B Flight-Test Progress Overview,” 47th AIAAAerospace Sciences Meeting and Exhibit , Orlando, FL, January 2009.

3McCullers, L. A., FLOPS Flight Optimization System, FLOPS Users Manual, December 2009.4Gern, F. H., “Improved Aerodynamic Analysis for Hybrid Wing Body Concept Design Optimization,” 50th AIAA

Aerospace Sciences Meeting and Exhibit , Nashville, TN, January 2012.5Gern, F. H., “Finite Element Based HWB Centerbody Structural Optimization and Weight Prediction,” 53rd

AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, Honolulu, HI, April 2012.6Gern, F. H., “Update on HCDstruct–A Tool for Hybrid Wing Body Conceptual Design and Structural Optimization,”

15th AIAA Aviation Technology, Integration, and Operations Conference, Dallas, TX, June 2015.7Gern, F. H., “Conceptual Design and Structural Analysis of an Open Rotor Hybrid Wing Body Aircraft,” 54th

AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference, Boston, MASS, April 2013.8Chhabra, R., Mulani, S. B., Kapania, R. K., and Schetz, J. A., “Control Power Optimization using Artifical Intelligence

for Hybrid Wing Body Aircraft,” 16th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Dallas, TX,June 2015.

9Nigam, N., Ayyalasomayajula, S., Qi, X., and Chen, P. C., “High-Fidelity Weight Estimation for Aircraft ConceptualDesign Optimization,” 16th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Dallas, TX, June 2015.

10Quinlan, J. R. and Gern, F. H., “Conceptual Design and Structural Optimization of ERA Hybrid Wing Body Concepts,”AIAA SciTech Forum and Exposition, San Diego, CA, January 2016.

11Hahn, A., “Vehicle Sketch Pad: A Parametric Geometry Modeler for Conceptual Aircraft Design,” 48th AIAA AerospaceSciences Meeting and Exhibit , Orlando, FL, January 2010.

12MSC Nastran 2008, Quick Reference Guide, MSC Software Corporation, Santa Ana, CA, 2008.13Velicki, A., “Damage Arresting Composites for Shaped Vehicles,” Tech. rep., NASA/CR-2009-215932, NASA Langley

Research Center, Hampton, VA, September 2009.14LeDoux, S. T., Herling, W. W., Fatta, G. J., and Ratcliff, R. R., “MDOPT–A Multidisciplinary Design Optimization

System using Higher Order Analysis Codes,” 10th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference,Albany, NY, September 2004.

12 of 13


15Gern, F. H., Vicroy, D., Mulani, S. B., Chhabra, R., Kapania, R. K., Schetz, J. A., Brown, D., and Princen, N. H.,“Artificial Intelligence Based Control Power Optimization on Tailless Aircraft,” Tech. rep., NASA/TM-2014-218671, NASALangley Research Center, Hampton, VA, December 2014.

16McKay, M. D., Beckman, R. J., and Conover, W. J., “Comparison of Three Methods for Selecting Values of InputVariables in the Analysis of Output from a Computer Code,” Technometrics, Vol. 21, No. 2, 1979, pp. 239–245.

13 of 13


Documents

Optimization of an Advanced Hybrid Wing Body Concept ......Optimization of an Advanced Hybrid Wing Body Concept using HCDstruct Version 1.2 Jesse R. Quinlan and Frank H. Gerny NASA